Document Type

Book Chapter

Publication Date



As discussed in previous chapters, artificial intelligence (AI) has the potential to be involved in almost all aspects of the health care industry. The legal landscape for health care AI is complex; AI systems with different intended uses, audiences, and use environments face different requirements at state, federal, and international levels. A full accounting of these legal requirements, or of the policy questions involved, is far beyond the scope of this chapter. Additionally, the legal and regulatory framework for AI in health care continues to evolve, given the nascent stage of the industry.

In this chapter, we offer an overview of the landscape through early April 2019 and undertake three tasks. First, we lay out a broad overview of laws applicable to different forms of health care AI, including federal statutes, federal regulations, and state tort law liability. Second, we address in considerable depth the regulatory requirements imposed on AI systems that help inform or make decisions about individual patients, such as diagnosis or treatment recommendations; these systems are referred to in this report as clinical AI. Clinical AI faces the closest scrutiny, especially by the U.S. Food and Drug Administration (FDA) and by other regulatory agencies internationally. These systems must demonstrate safety and efficacy. They may also generate liability under state tort law, which performs its own regulatory role and is intimately tied to the way FDA oversees clinical AI systems. Third, we note the legal and policy issues around privacy and patient data that affect clinical AI as well as other health care AI systems. Throughout the chapter, we highlight key challenges, opportunities, and gaps in the current framework. The chapter concludes with key considerations for addressing some of these issues.


Copyright 2022 by the National Academy of Sciences. All rights reserved. Reproduced with permission.